Nvidia targets robotaxi launch by 2027, signalling push beyond AI chips

Chipmaker bets on autonomous fleets and software as it looks for growth beyond data centres.

• Nvidia plans to test a Level 4 robotaxi service with a partner by 2027
• Service to launch with limited availability; partner and city undisclosed
• Nvidia aims to expand beyond AI chips into autonomous vehicles and robotics

Did our AI summary help?

Nvidia on Monday said it plans to test a robotaxi service with a partner as early as 2027, a move that underscores the chipmaker’s ambition to become a major player in self-driving vehicles, according to a report by CNBC.

The proposed service would use Level 4 autonomous driving, meaning vehicles can operate without human intervention within defined areas. Nvidia has not disclosed the partner or the city where the service would launch, but said it would start with limited availability.

“We will probably start with a limited availability but work with the partner for us to get our footing,” Xinzhou Wu, Nvidia’s vice president of automotive, said at a recent self-driving demonstration in San Francisco, as cited by CNBC.

A new growth lane beyond AI infrastructure

Autonomous driving and robotics are among Nvidia’s few major growth bets outside its core AI infrastructure business. Automotive and robotics chips generated $592 million in revenue in the quarter ended October, about 1% of Nvidia’s total sales, highlighting how early-stage the segment remains.

Chief Executive Jensen Huang has repeatedly said robotics, including self-driving cars, is Nvidia’s second-most important growth category after artificial intelligence.

At the CES conference in Las Vegas on Monday, Huang said the long-term vision is far larger than driver-assist features. “We imagine that someday, a billion cars on the road will all be autonomous,” he said, according to CNBC.

From car chips to full fleets

Nvidia has supplied automotive chips under its Drive brand since 2015. More recently, it has expanded into software, simulation tools and cloud access to its AI chips, allowing carmakers to train self-driving models and cut research and development costs.

The company says its Drive AGX Thor automotive computer, priced at about $3,500 per chip, helps manufacturers accelerate time to market by bundling computing power with Nvidia’s software stack.

Carmakers can also fine-tune the system for their vehicles, including how aggressively a car accelerates or brakes, Nvidia executives said.

Mercedes-Benz tie-up and road testing

Nvidia’s ambitions are already visible in partnerships. The company said in December that Mercedes-Benz models due in late 2026 will be able to use Nvidia’s technology to navigate complex urban environments such as San Francisco.

In December, Nvidia gave reporters an hour-long ride in a 2026 Mercedes-Benz CLA sedan in San Francisco. While a safety driver remained behind the wheel, the car drove itself for most of the trip, CNBC reported. The system struggled only in a dense traffic bottleneck involving buses, parked trucks and a competing autonomous vehicle.

Nvidia described the demo as “Level 2 Plus Plus”, similar to Tesla’s Full Self-Driving, meaning the driver must remain alert and ready to intervene.

Robotaxis enter a crowded field

Robotaxi services have gained traction over the past year, led by Waymo, which already operates commercial driverless taxis in five US cities, including San Francisco. Nvidia’s move signals that it wants exposure not just to consumer vehicles, but also to autonomous fleets — a business model that could generate recurring software and computing revenue.

The company is targeting 2028 for point-to-point self-driving features in consumer cars, betting that recent advances in generative AI will allow vehicles to handle more complex driving scenarios.